A torso position and rotation tracking system targeted towards seated or “standing in place” VR applications. More specifically, to enable leaning and limited rotation inside of a VR environment and to provide an in game connection point between the head tracking of an Oculus Rift and the absolute tracking of a Razor Hydra.
Design Goals
1) “good enough” position tracking to add to the Oculus Rift development kit.
2) Inexpensive.
3) Relatively simple and straight forward hardware.
How will it work?
Hardware - Four LEDs are placed onto a rigid “T” or cross shaped structure which is then worn on the users back; the LEDs are then tracked via a camera. This tracking concept has been clearly demonstrated many times (Johnny Chung Lee, et al).
Software - An extension of the Oculus “neck modeling” to include “torso modeling”.
Why is this worn on the back?
With any optical (camera based) system, one of the big issues is occlusion, which is when the camera doing the tracking can no longer see the points it is attempting to track. A tracker worn on the front of your torso can easily be blocked when bending forward to interact with a virtual object, when crouching down, or while using an arm/hand tracking device such as the Razor Hydra, Leap Motion, etc. A tracker worn on the back of your torso doesn’t share these restrictions.
In addition, the real world movements of an individual’s torso are much more limited and less precise than head movements. Attempting positional tracking via markers on an individual’s head means you first must go through the process of duplicating fairly accurate head tracking and then work out the body position afterwards based on the determined head pose. Since it’s a fair assumption that an individual’s head will be attached to a torso, tracking the torso directly should be a less complex way to achieve good results for head positioning.
Usage assumptions
1) User has an HMD and head tracking method is already established.
2) Seated or standing in place VR. (seated VR requires low back chair or stool)
3) < 180 degree user torso rotation
4) ~6ft of clear space behind the user to mount camera.
5) The tracking structure should be attached/positioned so as to not be affected by independent shoulder movement.
How expensive will it be to make one?
The hardware needed to build such a device can generally be purchased for <$50 USD.
Two potential hardware solutions are:
• Inferred LEDs and a Wii-mote (advantages: built-in image processing, wireless).
• Visible LEDS and a PS3 Eye (advantages: higher native res, no batteries, no Bluetooth syncing).
Also needed:
1) An adjustable, form fitting vest or harness.
2) A rigid mounting surface for the precise placement of LEDs. (Aluminum, MDF, plastic)
3) The tracking LEDs need power; batteries would be best, but they could also be wired (USB, AC adaptor).
Additional thoughts on rotation
By default, the system would only be used to rotate the in game torso, but not actually change the in game heading.
360 degree in-game rotation could be possible, but would need to be handled through some adapted control mechanism:
- a) Rotation being mapped in the same manner as a gamepad’s right thumbstick.
b) Rotating away from center is scaled greater than rotating towards center.